Original Research Stability Monitoring of the Nitrification Process: Multivariate Multivariate Statistical

  • View
    0

  • Download
    0

Embed Size (px)

Text of Original Research Stability Monitoring of the Nitrification Process: Multivariate Multivariate...

  • Introduction

    An effective and stable biological nitrification process requires appropriate selection of technological parameters different for each wastewater treatment plant due to its own operating conditions. The working conditions of the activated sludge should be properly selected in order to develop a large and diverse population of nitrifying bacteria. Growth rate of

    the nitrifying bacteria mainly depends on substrate availability, i.e., NH4

    + and NO2 − ions and concentrations

    of dissolved oxygen DO [1]. The first stage of nitrification is carried out by

    the bacteria Nitrosomonas, Nitrosococcus, and Nitrosospira (ammonia-oxidizing bacteria, AOB) which oxidize ammonium ions via an intermediate hydroxylamine (AOB1) to nitrite ions (AOB2) [2-4]. AOB are characterized by a long growth rate in comparison with heterotrophic bacteria, and sensitivity to toxic substances, i.e., nitrite ions. At the second stage of nitrification bacteria Nitrobacter, Nitrococcus, and Nitrospira (nitrite oxidizing bacteria, NOB) oxidize

    Pol. J. Environ. Stud. Vol. 27, No. 5 (2018), 2303-2313

    Original Research

    Stability Monitoring of the Nitrification Process: Multivariate Statistical Analysis

    Ewa Wąsik*, Krzysztof Chmielowski, Agnieszka Cupak, Grzegorz Kaczor

    University of Agriculture in Kraków, Department of Sanitary Engineering and Water Management, Kraków, Poland

    Received: 9 August 2017 Accepted: 24 September 2017

    Abstract

    The aim of this article is to define the possibilities of applying multivariate statistical analysis (PCA and control charts) in the monitoring of the effectiveness of biological nitrification in a wastewater treatment plant working for the municipality of Sanok. The difference in oxygen affinity between ammonium and nitrite oxidizers results in a bacteria competition between AOM and NOM. A more stable nitrification process was obtained in reactor I for mean oxygen concentration of 1.13-2.05 mgO2·dm

    -3. The lowest mean concentrations of ammonia nitrogen were obtained in the range 3.43-3.62 mgN-NH4

    +·dm-3. Reactor II worked at mean oxygen concentration 1.69-4.56 mgO2·dm -3,

    which caused lower stability in this study period. The mean concentration of ammonium nitrogen ranged from 4.06 to 9.08 mgN-NH4

    +·dm-3. April 2016 was considered the most stable period of work of nitrification reactors. In that month, in reactor I the upper specification limit USL was not exceeded. In reactor II the USL was exceeded only 10% of the time. The index of the process capacity Cpk was higher for reactor I, and achieved a value of 1.71. The process of nitrification in both reactors was qualified as stable when oxygen concentration was between 1 and 2 mgO2·dm

    -3.

    Keywords: multivariate statistical analysis, control chart, wastewater treatment

    *e-mail: ewa.wasik@urk.edu.pl

    DOI: 10.15244/pjoes/77958 ONLINE PUBLICATION DATE: 2018-04-13

  • 2304 Wąsik E., et al.

    nitrite ions formed at the first stage to nitrate ions [2-4]. This stage depends on nitrite ions formed by AOB and concentrations of dissolved oxygen.

    At low oxygen concentrations (0.3 mgO2·dm -3), AOB1

    are predominantly caused by the lowest oxygen affinity constant. At the oxygen concentrations within limits from 0.6 to 1.0 mgO2·dm

    -3, AOB2 wins the competition because of its higher maximum growth rate [5-6]. At the DO concentrations of 1.2-1.5 mgO2·dm

    -3 the difference in oxygen affinity between ammonium and nitrite oxidizers results in a competition between predominant AOB and NOB [7]. This caused the accumulation of toxic nitrite or the formation of toxic by-products such as NO and N2O [2-3, 8]. Literature reports that nitrite oxidation could be inhibited below 4.0 mgO2·dm

    -3 [9]. For relatively high reactor oxygen concentrations and not too low influent ammonium concentrations, Volckle et al. [6] found the occurrence operating zones AOB1+NOB and AOB2+NOB, resulting in nitrate formation.

    The use of electricity for aeration in wastewater treatment plants (WWTP) accounts for 50-90% of total electricity consumption, which translates into up to 30% of total operating costs [10]. Excessive oxygen in the nitrification reactor usually results in worse sedimentation of the flocs in the sludge tank. It can also cause problems with the denitrification process, due to the oxygen returned to the non-air reactors with recirculation streams. Most WWTPs are operated at dissolved oxygen concentration in limits 2-2.5 mgO2·dm

    -3 in order to ensure complete nitrification of ammonia to nitrate and the establishment of stable populations of bacteria AOB and NOB. This value 2 mgO2·dm

    -3 is popular lower limit in reactors with bubble aeration [3]. The optimal dissolved oxygen concentration is related to an ammonium load. In order to term the optimal conditions for the nitrification process it should be checked if nitrification goes properly with lower oxygen concentrations of 1 mgO2·dm

    -3. This value defines the concentration at which the process of nitrification starts running with lower speed [11]. A reduction in oxygen concentration from 2.5 to 1 mgO2·dm

    -3 decreases the unnecessary consumption of electricity.

    Information about oxygen concentration will help us better understand the biochemical processes by the WWTP operators and determine the optimal parameters for aeration. The amount of required oxygen should be tested in real conditions based on the measurements from online sensors of various nitrogen forms (ammonium, nitrites, nitrates). Online measurements and the application of a steering system and automation of SCADA (supervisory control and data acquisition) type allow for quicker detection of incorrect work or the failure in measurement instruments and sensors in a wastewater treatment plant, and consequently increase the technical reliability of the object [11]. To make steering and monitoring of technological processes more efficient, special software is applied and the management of vast installations is carried out

    with a computer in the scheduler’s room. Apart from the system of information on the required service of the instruments (e.g., in the situation of failures), the course of technological processes should be enriched by the systems of its visualization with the possibility of their statistical analysis. For this purpose, specialist programs (i.e., ASIM, BioWin, and SIMBA) can be used. Their computer simulations describe the real course of technological processes [11]. Another easily accessible tool for the analysis and visualization of data in Microsoft Windows is Excel.

    The observations of the studied process, i.e., according to the definition of the ordered sequence of changes, taking place subsequently in subsequent periods of times (e.g., hours, days, months), are often presented graphically using time series. One can single out the trend in cyclic or seasonal fluctuations due to the influence of various factors on the given phenomenon [12]. However, in such a big set of numbers, it is difficult to analyze and detect non-random changes (special disturbances) resulting from circumstances independent of the variability of the observed parameter of the process (affecting the process from outside). In quick detection of such disturbances, which can affect the process, and consequently, deteriorate the quality of the process, statistical process control (SPC) was applied. Usually it is carried out with Shewhart’s control charts. This method was monitored and the regulation was proposed as graphical procedure, in which the main role is fulfilled by a properly prepared diagram (chart). Dr Walter Shewhart [13] explained that the process is regarded as controlled, when the experience from the past allows us to predict (at least approximately) the probability that the observed variables are within certain borders. The construction of a control chart is not very complicated, but it does require prior organization. For this purpose, it is necessary to find the frequency of the observation of variables (e.g., every 24 hours) and the number of observations (e.g., 24 times in 24 hours). Equally important is the instruments’ precision of measurement and recording data obtained from them. The last step is a correct statistical analysis of the obtained results, compliant with the assumed technological assumptions.

    In the control charts the technological process can be visualized in time in such a way that subsequent observations are graphically presented on the abscissa. In the case when numerical data are present in subgroups of the same number (k), the most popular chart to monitor variables for the mean value of the process (Xmean) was applied. While making the graph, the value of the observed characteristic of the variable (i.e., in this case the mean) was given on the ordinate. Additionally, apart from the central line (CL), a typical control chart contains two control limits: the lower control limit (LCL) and upper control limit (UCL). Control limits are established based on the variability inside the subgroup by the calculation of standard deviation. Usually single points corresponding to the

  • 2305Stability Monitoring of the Nitrification...

    values calculated from the variable are connected with lines. In cases when such a line exceeds the upper or lower control line or the line has an unusual appearance or some systematic layout, one can say that the process became deregulated [14-16].

    The normal distr